The invention belongs to the technical field of automatic driving, and discloses a hierarchical decoupling method for an automatic driving city scene, which comprises the following steps of: 1, unifying static scene elements in a test scene to a traffic infrastructure coordinate system, and obtaining a grid map by using a probability placeholder method; 2, unifying vehicle-mounted multi-sensor data to a vehicle body coordinate system, and predicting and establishing a time sequence diagram by using a Kalman filter; step 3, based on the time sequence diagram, constructing an attention model by using a convolutional neural network and extracting traffic flow features; step 4, decoupling the nonlinear system through a decoupling pre-compensator, and then outputting element features after decoupling; and step 5, performing decoupling based on the corrected decoupling pre-compensator to obtain element features, completing clustering of the evaluation elements, and extracting the evaluation elements. Static scene elements, sensor data and a multivariable control system of a vehicle individual level are combined, and the confidence coefficient in the test is improved.
本发明属于属于自动驾驶技术领域,公开了面向自动驾驶城市场景的分层解耦方法,包括:步骤一、将测试场景中的静态场景要素统一至交通基础设施坐标系下,并利用概率占位方法得到栅格地图;步骤二、将车载多传感器数据统一至车体坐标系下,再利用卡尔曼滤波器预测建立时序图;步骤三、基于时序图,利用卷积神经网络构建注意力模型并提取交通流特征;步骤四、通过解耦预补偿器对所述非线性系统进行解耦,再输出解耦后的要素特征;步骤五、基于修正后的解耦预补偿器解耦获得要素特征完成测评要素的聚类,提取测评要素。本发明将静态场景要素、传感器数据和车辆个体层面的多变量控制系统相结合,提升了测试中的置信度。
Layered decoupling method for automatic driving city scene
面向自动驾驶城市场景的分层解耦方法
2024-11-08
Patent
Electronic Resource
Chinese
Driving scene restoration method and system for automatic driving vehicle
European Patent Office | 2022
|Automatic driving control method and device applied to driving test scene
European Patent Office | 2025
|